Dear Statalist,
I am currently trying to perform a -clogit- on a stratified
sample. The stratification is both choice-based and exogenous, as
the persons running the survey wanted to have the same amount of
persons for each *options* and for each *area*.
For example, suppose that in real life there are
10 persons in Region A using option 1
20 persons in Region A using option 2
30 persons in Region A using option 3
150 persons in Region B using option 1
250 persons in Region B using option 2
350 persons in Region B using option 3
then my sample will look like this:
5 persons in Region A using option 1
5 persons in Region A using option 2
5 persons in Region A using option 3
5 persons in Region B using option 1
5 persons in Region B using option 2
5 persons in Region B using option 3
How would you correct the clogit estimates to take this into account?
I know from chapter 8 "Discrete Choice Analysis : Theory and
Application to Travel Demand" written in 1985 by Ben-Akiva & Lerman
that only the intercept will be affected in a choice-based sample
and that I can correct it easily. I am not sure what to do with the
exogeneous part though.
Can stata handle this?
Thanks,
Simon
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
